Synthetic aperture radar sea surface wind field inversion method considering Doppler centroid anomaly

Abstract

The invention discloses a synthetic aperture radar sea surface wind field inversion method considering Doppler centroid anomaly, and the method comprises the following steps: obtaining vertical polarization SAR image data, and extracting a radar backscattering coefficient NRCS and an original Doppler centroid anomaly DCA related image; based on Doppler frequency data of the VV polarized SAR image, through quintic polynomial fitting and land area correction processing, DCA information caused by geophysical factors is extracted; acquiring actually measured wind speed data of the buoy, matching the actually measured wind speed data with an external background wind field, and performing nonlinear fitting correction by adopting an exponential function to obtain a corrected background wind field speed; constructing a variational analysis model containing an NRCS and DCA observation field, taking the corrected external wind field as a background field, and obtaining a wind speed field and a wind direction field through minimum cost function inversion; performing verification analysis on an inversion result by adopting buoy observation data subjected to space-time matching; the precision and reliability of the vector parameters of the sea surface wind field can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind field inversion, and in particular to a synthetic aperture radar sea surface wind field inversion method taking Doppler centroid anomaly into consideration. Background Art

[0002] Ocean winds are a crucial component of atmospheric circulation and are crucial for global climate regulation, renewable energy production, and marine ecosystem protection. SAR (Spectral Anomaly Recognition) has a spatial resolution exceeding 1 km and is capable of acquiring wind information in complex environments such as closed bays, fjords, and inland waters. It has become the mainstream technology for high-resolution ocean surface wind inversion.

[0003] Currently, SAR-based wind vector inversion methods are mainly divided into direct inversion and variational analysis. The direct inversion method directly inverts sea surface wind speed based on the GMF. The wind direction information required by the GMF is usually provided by numerical weather prediction models or scatterometer measurements. The inversion is completed by finding the wind vector closest to the actual radar measurement. This method performs well in areas with small wind direction gradients at sea. However, in coastal areas, the coarse resolution of numerical weather prediction models or scatterometers (typically 0.25°) cannot capture small-scale wind direction variations caused by terrain effects, limiting the practicality of this method. Alternatively, wind direction can be approximated by wind streak features extracted from SAR images. However, this method has two major limitations: first, the formation of wind streaks mainly depends on thermal convection mechanisms, and wind streaks are not observed in all images; second, wind streaks are subject to 180° directional ambiguity, which requires the use of atmospheric models or external wind field data to eliminate this uncertainty. Summary of the Invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a synthetic aperture radar sea surface wind field inversion method taking into account Doppler centroid anomaly, comprehensively analyze the characteristic laws and correction methods of SAR image DCA, and combine with the background wind field correction algorithm to propose a sea surface wind field inversion algorithm that simultaneously considers SAR image NRCS and DCA information to solve the problems existing in the background technology.

[0005] Technical solution: The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly described in the present invention is characterized by comprising the following steps: S1, acquire vertically polarized SAR image data, extract radar backscatter coefficient NRCS and original Doppler centroid anomaly DCA correlation image; S2, based on the Doppler frequency data of VV polarimetric SAR images, extracts the DCA information caused by geophysical factors through fifth-order polynomial fitting and land area correction processing; S3, obtaining the measured wind speed data of the buoy and matching it with the external background wind field, and performing nonlinear fitting correction using an exponential function to obtain the corrected background wind field speed; S4, construct a variational analysis model including the NRCS and DCA observation fields, use the corrected external wind field as the background field, and obtain the wind speed and direction fields by minimizing the cost function; S5, use the time-space matched buoy observation data to verify and analyze the inversion results.

[0006] Furthermore, step S2 includes the following steps: S21, calculate the DCA value of the single-antenna SAR satellite echo signal : ; in, is the measured value of Doppler frequency, which indicates the frequency of the radar echo on the center line of the antenna beam; is the predicted value of Doppler frequency; S22, average each swath along the azimuth direction and use a quintic polynomial to perform least squares fitting: ; in, are coefficients determined by the least squares method; S23, remove the polynomial fitting value to extract the scallop signal : ; in, is the one-dimensional signal averaged in each swath direction, is the fifth-order polynomial fitted to the signal; S24, eliminating the scallop effect corresponding to each swath; S25, geophysical DCA is obtained by correcting the average profile of the land area distance : ; in, represents DCA after removing the scallop effect, The one-dimensional signal averaged over the range for each swath.

[0007] Furthermore, the cost function constructed in step S4 includes: ; in, Represents the observation item, which is composed of the observation data of SAR image; represents the background field term; the formula is as follows: ; in, is the wind speed, is the relative wind direction; represents the corrected background wind speed, Indicates the relative wind direction of the background wind field; and are the standard deviations of background wind speed and background wind direction, respectively.

[0008] Furthermore, the NRCS information observed by the SAR image is added to the observation field, which is expressed as: ; in, The geophysical model function CMOD is used to calculate the Predicted NRCS, is the NRCS obtained from SAR observations, is the standard deviation of the NRCS.

[0009] Furthermore, the DCA information observed by the SAR image is added to the observation field, which is expressed as: ; in, The geophysical model function CDOP is used to calculate the Predicted DCA information; represents the geophysical DCA after preprocessing; is the standard deviation of DCA.

[0010] Furthermore, the NRCS and DCA information observed by the SAR image are added to the observation field at the same time and expressed as: ; in, represents the cost function of adding the NRCS information observed by SAR images to the observation field; represents the cost function for adding DCA information observed from SAR images to the observation field; represents the background field term; is the wind speed, is the relative wind direction; The geophysical model function CMOD is used to calculate the predicted NRCS; is the NRCS obtained from SAR observations; is the standard deviation of the NRCS; The geophysical model function CDOP is used to calculate the Predicted DCA information; represents the geophysical DCA after preprocessing; is the standard deviation of DCA; Indicates the corrected background wind speed; Indicates the relative wind direction of the background wind field; and are the standard deviations of background wind speed and background wind direction, respectively.

[0011] Furthermore, the background wind field correction parameters in step S3 are determined by nonlinear least squares fitting.

[0012] Furthermore, the exponential function formula in step S3 is as follows: ; in, represents the background wind speed, represents the correction parameters obtained through nonlinear fitting; Indicates the corrected background wind speed.

[0013] An electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory, and the processor implements the steps of any one of the methods when executing the program.

[0014] The computer-readable storage medium of the present invention stores a computer program, which implements the steps of any one of the methods when executed by a processor.

[0015] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: when constructing the variational analysis method, the present invention innovatively integrates NRCS and DCA information at the same time; the present invention verifies that the background wind field wind speed error is the core factor restricting the performance of the variational analysis method, and corrects it through the least squares correction, which can effectively optimize the background wind field wind speed; compared with the wind speed and external wind field wind direction obtained by the direct inversion method, this method improves the inversion accuracy of wind speed by 10.6% and the inversion accuracy of wind direction by 4.5%, significantly improving the inversion accuracy of sea surface wind speed and wind direction, and providing a reliable method for high-precision evaluation of local wind fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of the present invention; Figure 2 The spatial coverage of the SAR image of the present invention and the buoy position used for wind inversion verification; Figure 3 The observed Doppler frequency (a), predicted Doppler frequency (b), uncorrected DCA (c), simulation of scallop effects for different swaths (d), DCA corrected for scallop terms (e), and DCA corrected for land use (f) are shown in the present invention. Figure 4Scatter plot comparison of the ERA5 wind speed of the present invention and the wind speed measured by the buoy (a), and the deviation of the ERA5 wind speed and the wind speed measured by the buoy before and after correction as the buoy wind speed changes (b); Figure 5 For the present invention Scatter plot comparison of the obtained wind speed and the wind speed measured by the buoy; Figure 6 For the present invention Comparison of the scatter plot of the obtained wind direction and the wind speed measured by the buoy; Figure 7 A scatter plot comparison of the wind speed obtained by the direct inversion method of the present invention and the wind speed measured by the buoy; Figure 8 A scatter plot comparison of the ERA5 wind direction of the present invention and the wind speed measured by the buoy. DETAILED DESCRIPTION

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0018] like Figure 1 As shown, an embodiment of the present invention provides a synthetic aperture radar sea surface wind field inversion method considering Doppler centroid anomaly, comprising the following steps: Step 1: Obtain vertical (VV) polarization SAR images and extract SAR radar backscatter coefficient (NRCS) and raw Doppler centroid anomaly (DCA) related images.

[0019] Specifically, this example uses the VV polarization L2 OCN (Ocean) product of the Sentinel-1 satellite launched by the European Space Agency as an example. This paper collects a total of 1803 scenes of Sentinel-1 OCN datasets in IW mode from 2015 to 2024. Its regional distribution is as follows: Figure 2 As shown. The Sentinel-1 OCN (Ocean) data product is designed to provide geophysical parameters related to wind, waves, and surface currents to a wide range of end users. It consists of three components: Ocean Wave Spectrum (OSW), Ocean Wind Index (OWI), and Radial Surface Velocity (RVL). This paper uses the NRCS (resolution approximately 1 km × 1 km) of the OWI component and the Doppler frequency observations of the RVL component. , predicted (geometric) Doppler frequency Wind vector inversion is performed at Doppler frequencies (resolution is approximately equal to 1 km × 1 km).

[0020] Step 2: Based on the Doppler frequency correlation data of the VV polarimetric SAR image obtained in step 1, the DCA information caused by geophysical factors is extracted.

[0021] Specifically, in step 2, taking a SAR image acquired in the Gulf of Mexico on April 30, 2020 as an example, the specific steps for extracting the geophysical DCA image based on the DCA-related data of the VV polarimetric SAR image are as follows: S21, calculate the DCA value of the single-antenna SAR satellite echo signal : ; in, is the measured value of the Doppler frequency (e.g. Figure 3 (a)), represents the frequency of the radar echo on the center line of the antenna beam; is the predicted value of Doppler frequency (e.g. Figure 3 (b)); It is mainly caused by the relative velocity between the SAR satellite and the Earth, which can be estimated by the precise orbit and antenna pointing of the SAR satellite. Indicates DCA (e.g. Figure 3 (c)); S22, for each swath Take the average value to obtain the one-dimensional signal profile in azimuth; then use the quintic polynomial least squares method to fit the signal profile to obtain the fitting value: ; in, are coefficients determined by the least squares method; S23, remove the polynomial fitting value to extract the periodic change signal, namely the scallop signal : ; in, The one-dimensional signal averaged in each swath direction (e.g. Figure 4 solid red line), is a fifth-order polynomial fitted to the signal (e.g. Figure 3 (d) blue solid line); S24, eliminate the scallop effect corresponding to each swath The results are as follows: Figure 3 (e) S25, geophysical DCA is obtained by correcting the average profile of the land area distance (like Figure 3 (f)): ; in, represents DCA after removing the scallop effect, The one-dimensional signal averaged over the range for each swath.

[0022] Step 3: Get the actual wind speed observation data of the buoy and match it with the external background wind field data, and then use the form The exponential function is nonlinearly fitted to obtain the corrected background wind speed. Where a represents the correction parameter obtained through nonlinear fitting.

[0023] Specifically, this paper uses actual wind speed observations from buoys at the U.S. National Data Buoy Center (NDBC) (https: / / www.ndbc.noaa.gov / ). Furthermore, the ERA5 reanalysis dataset (https: / / cds.climate.copernicus.eu / ) is used as the external wind direction and background wind field for the direct SAR wind field inversion method. The fifth-generation ERA5 reanalysis dataset, developed by the European Centre for Medium-Range Weather Forecasts (ECMWF), provides global climate and weather data from 1979 to the present, covering a wide range of atmospheric, oceanic, and land surface parameters. Its high temporal and spatial resolution—10 m altitude u and v component wind speed data on an hourly 0.25° × 0.25° grid (approximately 31 km spacing)—facilitates detailed analysis of short-term wind field variations.

[0024] This paper selects observation data from 27 buoys during 2015-2024 and matches them with ERA5 reanalysis data, obtaining a total of 5,206,485 valid data pairs ( Figure 4 (a)). The analysis results show that as the measured wind speed increases (greater than 3 m / s), the negative deviation between the ERA5 reanalysis data and the buoy observations gradually increases ( Figure 4 (b)). Therefore, in order to improve the accuracy of the background wind field, the present invention uses an exponential function for nonlinear fitting to correct the ERA5 wind speed. The corrected background wind field speed is obtained. It can be expressed as: ; like Figure 8 As shown in the figure, the absolute value of the overall deviation of the ERA5 data after correction has been significantly reduced from -0.54 to 0.01, indicating that systematic errors have been effectively controlled. In particular, the previously prevalent negative deviation phenomenon has been greatly improved when the wind speed is greater than 5 m / s.

[0025] In step 4, the NRCS and DCA information of the VV polarimetric SAR image is used as the observation field, and the corrected external wind field is used as the background field. The cost function is constructed for variational optimization to obtain the wind speed and direction fields inverted by the variational analysis method.

[0026] Specifically, in step 4, the NRCS and DCA information of the VV polarimetric SAR image is used as the observation field, and the corrected external wind field is used as the background field. The cost function is constructed for variational optimization to obtain the wind speed and direction fields inverted by the variational analysis method. The specific steps are as follows: Step 41, using the observation data of the VV polarimetric SAR image as the observation field and the external wind field as the background field, construct the cost function J: ; in, Represents the observation item, which is composed of the observation data of SAR image; Represents the background field term; considering that the errors of wind speed and wind direction in SAR observations are independent of each other, the observation field treats these two variables as independent quantities, and the formula is as follows: ; in, is the wind speed, is the relative wind direction; represents the corrected background wind speed, Indicates the relative wind direction of the background wind field; and are the standard deviations of background wind speed and background wind direction, which are 1.7 and 20 respectively.

[0027] Step 42: Construct the observation field considering NRCS The CMOD pattern function used for VV polarimetric SAR sea surface wind field inversion is often used as the input geophysical model function (GMF) for wind speed inversion. The CMOD pattern function is based on Bragg resonant scattering theory and is derived by establishing a semi-empirical geophysical model function between the NRCS and the radar incident angle, wind speed at 10 m above the sea surface, and relative wind direction. Its general form can be expressed as: ; in, represents VV polarized NRCS; is the wind speed at a height of 10 m above the sea surface; Represents the relative wind direction angle, which is defined as the angle between the actual wind direction and the azimuth of the radar antenna; is the radar incident angle. The dependence of the backscatter coefficient on wind speed and incident angle is characterized. 、 The modulating effect of the wind direction is described. For the CMOD5.N function, its parameters are fitted with the ECMWF wind field data under the equivalent neutral atmospheric boundary conditions, and the model function n is set to 1.6.

[0028] The NRCS information observed by SAR images is added to the observation field, which can be expressed as: ; in, The geophysical model function CMOD is used to calculate the Predicted NRCS, is the NRCS obtained from SAR observations, The standard deviation of NRCS is 0.5.

[0029] Step 43: Construct the observation field considering DCA :Geophysical DCA It can be expressed as: ; in, is the angle of incidence (in degrees), is the angle (in degrees) between the wind direction at 10 m and the antenna line of sight, is the wind speed at 10 m, Indicates the polarization mode. and is Both depend on the polarization coefficients.

[0030] Adding the DCA information observed by SAR images to the observation field can be expressed as: ; in, The geophysical model function CDOP is used to calculate the Predicted DCA information; represents the geophysical DCA after preprocessing; is the standard deviation of DCA.

[0031] Step 44, construct a variational analysis method VAM that comprehensively considers NRCS and DCA ( ): Adding the NRCS and DCA information observed by SAR images to the observation field at the same time can be expressed as: ; The total cost function obtained is: ; Step 45: Input the external wind field layer, VV polarization NRCS image and DCA image and radar incident angle information into In the process, the system iterates through the wind speed and direction parameter space to minimize , the determined global optimal solution corresponds to VAM ( ) The wind speed and direction obtained by inversion (such as Figure 5 and Figure 6shown).

[0032] Step 5: Verify and analyze the wind speed and direction results of the SAR inverted wind field using the spatiotemporally matched buoy observation data. Specifically, the present invention spatiotemporally matches 1803 scenes of Sentinel-1 OCN data in IW mode with the measured data from 27 NDBC buoys, obtaining a total of 2826 valid matching sample pairs. Among them, a single scene SAR image can be matched to a maximum of 6 buoy data, and most SAR images can only be matched to 1-2 buoy data (e.g. Figure 2 shown).

[0033] To compare and verify VAM ( ) performance, using the CMOD5.N geophysical model function and the wind direction of the ERA5 external wind field, the sea surface wind speed is inverted for the VV polarization NRCS image, and the wind speed field inverted by the direct inversion method is obtained (such as Figure 7 shown).

[0034] The results showed that VAM ( ) inversion has the best wind speed accuracy, with an RMSE of 1.35 m / s and a deviation of -0.06 m / s, which is better than the direct inversion method, with an RMSE of 1.51 m / s and a deviation of -0.03 m / s. ) also has the best wind direction accuracy, with an RMSE of 26.03 m / s and a deviation of -0.85 m / s, which is better than the external wind direction accuracy of ERA5 (such as Figure 8 As shown in Figure 2), its RMSE is 27.25 m / s and the deviation is 0.91 m / s. ) improves the inversion accuracy of wind speed by 10.6% and wind direction by 4.5%, indicating that the SAR sea surface wind field inversion method using radar backscattering and Doppler shift proposed in the present invention can improve the inversion accuracy of sea surface wind speed and wind direction, and provide a reliable method for high-precision evaluation of local wind fields.

[0035] In general, the present invention has important scientific value and practical significance in promoting the development of SAR remote sensing wind field inversion technology, expanding the application field of SAR image DCA, and improving the ability of marine environment monitoring.

Claims

1. A synthetic aperture radar sea surface wind field inversion method considering Doppler centroid anomaly, characterized by: The following steps are involved: S1, acquire vertically polarized SAR image data, extract radar backscatter coefficient NRCS and original Doppler centroid anomaly DCA correlation image; S2, based on the Doppler frequency data of VV polarimetric SAR images, extracts DCA information caused by geophysical factors through fifth-order polynomial fitting and land area correction processing; S3, obtaining the measured wind speed data of the buoy and matching it with the external background wind field, and performing nonlinear fitting correction using an exponential function to obtain the corrected background wind field speed; S4, construct a variational analysis model including the NRCS and DCA observation fields, use the corrected external wind field as the background field, and obtain the wind speed and direction fields by minimizing the cost function; S5, use the time-space matched buoy observation data to verify and analyze the inversion results.

2. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 1, characterized in that: Step S2 includes the following steps: S21, calculate the DCA value of the single-antenna SAR satellite echo signal : ； in, is the measured value of Doppler frequency, which indicates the frequency of the radar echo on the center line of the antenna beam; is the predicted value of Doppler frequency; S22, average each swath along the azimuth direction and use a quintic polynomial to perform least squares fitting: ； in, are the coefficients determined by the least squares method; S23, remove the polynomial fitting value to extract the scallop signal : ； in, is the one-dimensional signal averaged in each swath direction, is the fifth-order polynomial fitted to the signal; S24, eliminating the scallop effect corresponding to each swath; S25, geophysical DCA is obtained by correcting the average profile of the land area distance : ； in, represents DCA after removing the scallop effect, The one-dimensional signal averaged over the range for each swath.

3. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 2, characterized in that: The cost function constructed in step S4 includes: ； in, Represents the observation item, which is composed of the observation data of SAR image; represents the background field term; the formula is as follows: ； in, is the wind speed, is the relative wind direction; represents the corrected background wind speed, Indicates the relative wind direction of the background wind field; and are the standard deviations of background wind speed and background wind direction, respectively.

4. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 3, characterized in that: The NRCS information observed by the SAR image is added to the observation field and expressed as: ； in, The geophysical model function CMOD is used to calculate the Predicted NRCS, is the NRCS obtained from SAR observations, is the standard deviation of the NRCS.

5. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 4, characterized in that: The DCA information observed by the SAR image is added to the observation field, which is expressed as: ； in, The geophysical model function CDOP is used to calculate the Predicted DCA information; represents the geophysical DCA after preprocessing; is the standard deviation of DCA.

6. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 5, characterized in that: The NRCS and DCA information observed by SAR images are added to the observation field at the same time and expressed as: ； in, represents the cost function of adding the NRCS information observed by SAR images to the observation field; represents the cost function for adding DCA information observed from SAR images to the observation field; represents the background field term; is the wind speed, is the relative wind direction; The geophysical model function CMOD is used to calculate the predicted NRCS; is the NRCS obtained from SAR observations; is the standard deviation of the NRCS; The geophysical model function CDOP is used to calculate the Predicted DCA information; represents the geophysical DCA after preprocessing; is the standard deviation of DCA; Indicates the corrected background wind speed; Indicates the relative wind direction of the background wind field; and are the standard deviations of background wind speed and background wind direction, respectively.

7. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 1, characterized in that: The background wind field correction parameters in step S3 are determined by nonlinear least squares fitting.

8. The method for inverting sea surface wind field using synthetic aperture radar considering Doppler centroid anomaly according to claim 1, characterized in that: The exponential function formula in step S3 is as follows: ； in, represents the background wind speed, represents the correction parameters obtained through nonlinear fitting; Indicates the corrected background wind speed.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that: When the processor executes the program, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images